DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver
@ 2020-01-16  9:27 Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 1/4] net/hns3: replace memory barrier with data dependency order Wei Hu (Xavier)
                   ` (4 more replies)
  0 siblings, 5 replies; 6+ messages in thread
From: Wei Hu (Xavier) @ 2020-01-16  9:27 UTC (permalink / raw)
  To: dev

This series are improvement and cleanup for hns3 PMD driver.

Chengwen Feng (1):
  net/hns3: replace memory barrier with data dependency order

Lijun Ou (2):
  net/hns3: remove unnecessary input parameter valid judgement
  net/hns3: remove unnecessary branch

Wei Hu (Xavier) (1):
  doc: update release notes for hns3

 doc/guides/nics/hns3.rst               |  1 +
 doc/guides/rel_notes/release_20_02.rst |  6 ++
 drivers/net/hns3/hns3_ethdev.c         |  2 -
 drivers/net/hns3/hns3_flow.c           |  2 -
 drivers/net/hns3/hns3_rxtx.c           | 85 ++++++++++++++++++++++----
 5 files changed, 80 insertions(+), 16 deletions(-)

-- 
2.23.0


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [dpdk-dev] [PATCH 1/4] net/hns3: replace memory barrier with data dependency order
  2020-01-16  9:27 [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Wei Hu (Xavier)
@ 2020-01-16  9:27 ` Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 2/4] net/hns3: remove unnecessary input parameter valid judgement Wei Hu (Xavier)
                   ` (3 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: Wei Hu (Xavier) @ 2020-01-16  9:27 UTC (permalink / raw)
  To: dev

From: Chengwen Feng <fengchengwen@huawei.com>

This patch optimizes the Rx performance by using data dependency ordering
to instead of memory barrier which is rte_cio_rmb in the '.rx_pkt_burst'
ops implementation function named hns3_recv_pkts.

Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
---
 drivers/net/hns3/hns3_rxtx.c | 85 +++++++++++++++++++++++++++++++-----
 1 file changed, 73 insertions(+), 12 deletions(-)

diff --git a/drivers/net/hns3/hns3_rxtx.c b/drivers/net/hns3/hns3_rxtx.c
index 6f74a7917..9d8d0b7e1 100644
--- a/drivers/net/hns3/hns3_rxtx.c
+++ b/drivers/net/hns3/hns3_rxtx.c
@@ -1402,13 +1402,14 @@ hns3_rx_set_cksum_flag(struct rte_mbuf *rxm, uint64_t packet_type,
 uint16_t
 hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
 {
+	volatile struct hns3_desc *rx_ring;  /* RX ring (desc) */
+	volatile struct hns3_desc *rxdp;     /* pointer of the current desc */
 	struct hns3_rx_queue *rxq;      /* RX queue */
-	struct hns3_desc *rx_ring;      /* RX ring (desc) */
 	struct hns3_entry *sw_ring;
 	struct hns3_entry *rxe;
-	struct hns3_desc *rxdp;         /* pointer of the current desc */
 	struct rte_mbuf *first_seg;
 	struct rte_mbuf *last_seg;
+	struct hns3_desc rxd;
 	struct rte_mbuf *nmb;           /* pointer of the new mbuf */
 	struct rte_mbuf *rxm;
 	struct rte_eth_dev *dev;
@@ -1440,6 +1441,67 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
 		bd_base_info = rte_le_to_cpu_32(rxdp->rx.bd_base_info);
 		if (unlikely(!hns3_get_bit(bd_base_info, HNS3_RXD_VLD_B)))
 			break;
+		/*
+		 * The interactive process between software and hardware of
+		 * receiving a new packet in hns3 network engine:
+		 * 1. Hardware network engine firstly writes the packet content
+		 *    to the memory pointed by the 'addr' field of the Rx Buffer
+		 *    Descriptor, secondly fills the result of parsing the
+		 *    packet include the valid field into the Rx Buffer
+		 *    Descriptor in one write operation.
+		 * 2. Driver reads the Rx BD's valid field in the loop to check
+		 *    whether it's valid, if valid then assign a new address to
+		 *    the addr field, clear the valid field, get the other
+		 *    information of the packet by parsing Rx BD's other fields,
+		 *    finally write back the number of Rx BDs processed by the
+		 *    driver to the HNS3_RING_RX_HEAD_REG register to inform
+		 *    hardware.
+		 * In the above process, the ordering is very important. We must
+		 * make sure that CPU read Rx BD's other fields only after the
+		 * Rx BD is valid.
+		 *
+		 * There are two type of re-ordering: compiler re-ordering and
+		 * CPU re-ordering under the ARMv8 architecture.
+		 * 1. we use volatile to deal with compiler re-ordering, so you
+		 *    can see that rx_ring/rxdp defined with volatile.
+		 * 2. we commonly use memory barrier to deal with CPU
+		 *    re-ordering, but the cost is high.
+		 *
+		 * In order to solve the high cost of using memory barrier, we
+		 * use the data dependency order under the ARMv8 architecture,
+		 * for exmple:
+		 *      instr01: load A
+		 *      instr02: load B <- A
+		 * the instr02 will always execute after instr01.
+		 *
+		 * To construct the data dependency ordering, we use the
+		 * following assignment:
+		 *      rxd = rxdp[(bd_base_info & (1u << HNS3_RXD_VLD_B)) -
+		 *                 (1u<<HNS3_RXD_VLD_B)]
+		 * Using gcc compiler under the ARMv8 architecture, the related
+		 * assembly code example as follows:
+		 * note: (1u << HNS3_RXD_VLD_B) equal 0x10
+		 *      instr01: ldr w26, [x22, #28]  --read bd_base_info
+		 *      instr02: and w0, w26, #0x10   --calc bd_base_info & 0x10
+		 *      instr03: sub w0, w0, #0x10    --calc (bd_base_info &
+		 *                                            0x10) - 0x10
+		 *      instr04: add x0, x22, x0, lsl #5 --calc copy source addr
+		 *      instr05: ldp x2, x3, [x0]
+		 *      instr06: stp x2, x3, [x29, #256] --copy BD's [0 ~ 15]B
+		 *      instr07: ldp x4, x5, [x0, #16]
+		 *      instr08: stp x4, x5, [x29, #272] --copy BD's [16 ~ 31]B
+		 * the instr05~08 depend on x0's value, x0 depent on w26's
+		 * value, the w26 is the bd_base_info, this form the data
+		 * dependency ordering.
+		 * note: if BD is valid, (bd_base_info & (1u<<HNS3_RXD_VLD_B)) -
+		 *       (1u<<HNS3_RXD_VLD_B) will always zero, so the
+		 *       assignment is correct.
+		 *
+		 * So we use the data dependency ordering instead of memory
+		 * barrier to improve receive performance.
+		 */
+		rxd = rxdp[(bd_base_info & (1u << HNS3_RXD_VLD_B)) -
+			   (1u << HNS3_RXD_VLD_B)];
 
 		nmb = rte_mbuf_raw_alloc(rxq->mb_pool);
 		if (unlikely(nmb == NULL)) {
@@ -1463,14 +1525,13 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
 		rxe->mbuf = nmb;
 
 		dma_addr = rte_cpu_to_le_64(rte_mbuf_data_iova_default(nmb));
-		rxdp->addr = dma_addr;
 		rxdp->rx.bd_base_info = 0;
+		rxdp->addr = dma_addr;
 
-		rte_cio_rmb();
 		/* Load remained descriptor data and extract necessary fields */
-		data_len = (uint16_t)(rte_le_to_cpu_16(rxdp->rx.size));
-		l234_info = rte_le_to_cpu_32(rxdp->rx.l234_info);
-		ol_info = rte_le_to_cpu_32(rxdp->rx.ol_info);
+		data_len = (uint16_t)(rte_le_to_cpu_16(rxd.rx.size));
+		l234_info = rte_le_to_cpu_32(rxd.rx.l234_info);
+		ol_info = rte_le_to_cpu_32(rxd.rx.ol_info);
 
 		if (first_seg == NULL) {
 			first_seg = rxm;
@@ -1489,14 +1550,14 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
 		}
 
 		/* The last buffer of the received packet */
-		pkt_len = (uint16_t)(rte_le_to_cpu_16(rxdp->rx.pkt_len));
+		pkt_len = (uint16_t)(rte_le_to_cpu_16(rxd.rx.pkt_len));
 		first_seg->pkt_len = pkt_len;
 		first_seg->port = rxq->port_id;
-		first_seg->hash.rss = rte_le_to_cpu_32(rxdp->rx.rss_hash);
+		first_seg->hash.rss = rte_le_to_cpu_32(rxd.rx.rss_hash);
 		first_seg->ol_flags |= PKT_RX_RSS_HASH;
 		if (unlikely(hns3_get_bit(bd_base_info, HNS3_RXD_LUM_B))) {
 			first_seg->hash.fdir.hi =
-				rte_le_to_cpu_32(rxdp->rx.fd_id);
+				rte_le_to_cpu_32(rxd.rx.fd_id);
 			first_seg->ol_flags |= PKT_RX_FDIR | PKT_RX_FDIR_ID;
 		}
 		rxm->next = NULL;
@@ -1513,9 +1574,9 @@ hns3_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
 			hns3_rx_set_cksum_flag(rxm, first_seg->packet_type,
 					       cksum_err);
 
-		first_seg->vlan_tci = rte_le_to_cpu_16(rxdp->rx.vlan_tag);
+		first_seg->vlan_tci = rte_le_to_cpu_16(rxd.rx.vlan_tag);
 		first_seg->vlan_tci_outer =
-			rte_le_to_cpu_16(rxdp->rx.ot_vlan_tag);
+			rte_le_to_cpu_16(rxd.rx.ot_vlan_tag);
 		rx_pkts[nb_rx++] = first_seg;
 		first_seg = NULL;
 		continue;
-- 
2.23.0


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [dpdk-dev] [PATCH 2/4] net/hns3: remove unnecessary input parameter valid judgement
  2020-01-16  9:27 [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 1/4] net/hns3: replace memory barrier with data dependency order Wei Hu (Xavier)
@ 2020-01-16  9:27 ` Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 3/4] net/hns3: remove unnecessary branch Wei Hu (Xavier)
                   ` (2 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: Wei Hu (Xavier) @ 2020-01-16  9:27 UTC (permalink / raw)
  To: dev

From: Lijun Ou <oulijun@huawei.com>

Since the input parameter named dev has been guaranteed not to be NULL
in the rte layer of DPDK framework, it doesn't need to check dev whether
is null in the 'filter_ctrl' ops implementation function named
hns3_dev_filter_ctrl.

Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
---
 drivers/net/hns3/hns3_flow.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/net/hns3/hns3_flow.c b/drivers/net/hns3/hns3_flow.c
index aa614175d..98affa567 100644
--- a/drivers/net/hns3/hns3_flow.c
+++ b/drivers/net/hns3/hns3_flow.c
@@ -1887,8 +1887,6 @@ hns3_dev_filter_ctrl(struct rte_eth_dev *dev, enum rte_filter_type filter_type,
 	struct hns3_hw *hw;
 	int ret = 0;
 
-	if (dev == NULL)
-		return -EINVAL;
 	hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	switch (filter_type) {
 	case RTE_ETH_FILTER_GENERIC:
-- 
2.23.0


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [dpdk-dev] [PATCH 3/4] net/hns3: remove unnecessary branch
  2020-01-16  9:27 [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 1/4] net/hns3: replace memory barrier with data dependency order Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 2/4] net/hns3: remove unnecessary input parameter valid judgement Wei Hu (Xavier)
@ 2020-01-16  9:27 ` Wei Hu (Xavier)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 4/4] doc: update release notes for hns3 Wei Hu (Xavier)
  2020-01-17 10:15 ` [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Ferruh Yigit
  4 siblings, 0 replies; 6+ messages in thread
From: Wei Hu (Xavier) @ 2020-01-16  9:27 UTC (permalink / raw)
  To: dev

From: Lijun Ou <oulijun@huawei.com>

Because the rte layer of DPDK framework has already processed the case
when the idx is zero before calling the '.mac_addr_remove' ops function,
the input parameter named idx can not be zero in the '.mac_addr_remove'
function. This patch removes unnecessary branch process to check whether
input parameter named idx is zero in the '.mac_addr_remove' ops
implementation function named hns3_remove_mac_addr.

Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
---
 drivers/net/hns3/hns3_ethdev.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/net/hns3/hns3_ethdev.c b/drivers/net/hns3/hns3_ethdev.c
index 9866d147b..918fbe076 100644
--- a/drivers/net/hns3/hns3_ethdev.c
+++ b/drivers/net/hns3/hns3_ethdev.c
@@ -1473,8 +1473,6 @@ hns3_remove_mac_addr(struct rte_eth_dev *dev, uint32_t idx)
 		return;
 	}
 
-	if (idx == 0)
-		hw->mac.default_addr_setted = false;
 	rte_spinlock_unlock(&hw->lock);
 }
 
-- 
2.23.0


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [dpdk-dev] [PATCH 4/4] doc: update release notes for hns3
  2020-01-16  9:27 [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Wei Hu (Xavier)
                   ` (2 preceding siblings ...)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 3/4] net/hns3: remove unnecessary branch Wei Hu (Xavier)
@ 2020-01-16  9:27 ` Wei Hu (Xavier)
  2020-01-17 10:15 ` [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Ferruh Yigit
  4 siblings, 0 replies; 6+ messages in thread
From: Wei Hu (Xavier) @ 2020-01-16  9:27 UTC (permalink / raw)
  To: dev

From: "Wei Hu (Xavier)" <xavier.huwei@huawei.com>

Update release doc briefly describing updates to hns3 PMD for
20.02 release.

The support for these was added include the following commits:
1) commit 6e79a0ad1236 ("net/hns3: support Rx interrupt")
2) commit 34c5c2d5abb8 ("net/hns3: support setting VF MAC address by PF
                         driver")

Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
---
 doc/guides/nics/hns3.rst               | 1 +
 doc/guides/rel_notes/release_20_02.rst | 6 ++++++
 2 files changed, 7 insertions(+)

diff --git a/doc/guides/nics/hns3.rst b/doc/guides/nics/hns3.rst
index 505488b6c..8d19f4851 100644
--- a/doc/guides/nics/hns3.rst
+++ b/doc/guides/nics/hns3.rst
@@ -22,6 +22,7 @@ Features of the HNS3 PMD are:
 - Port hardware statistics
 - Jumbo frames
 - Link state information
+- Interrupt mode for RX
 - VLAN stripping
 - NUMA support
 
diff --git a/doc/guides/rel_notes/release_20_02.rst b/doc/guides/rel_notes/release_20_02.rst
index 1e83e659a..b69f45046 100644
--- a/doc/guides/rel_notes/release_20_02.rst
+++ b/doc/guides/rel_notes/release_20_02.rst
@@ -76,6 +76,12 @@ New Features
   Added support for L2TPv3 over IP rte_flow patterns to the testpmd
   application.
 
+* **Updated Hisilicon hns3 driver.**
+
+  Updated Hisilicon hns3 driver with new features and improvements, including:
+
+  * Added support for Rx interrupt.
+  * Added support setting VF MAC address by PF driver.
 
 Removed Items
 -------------
-- 
2.23.0


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver
  2020-01-16  9:27 [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Wei Hu (Xavier)
                   ` (3 preceding siblings ...)
  2020-01-16  9:27 ` [dpdk-dev] [PATCH 4/4] doc: update release notes for hns3 Wei Hu (Xavier)
@ 2020-01-17 10:15 ` Ferruh Yigit
  4 siblings, 0 replies; 6+ messages in thread
From: Ferruh Yigit @ 2020-01-17 10:15 UTC (permalink / raw)
  To: Wei Hu (Xavier), dev

On 1/16/2020 9:27 AM, Wei Hu (Xavier) wrote:
> This series are improvement and cleanup for hns3 PMD driver.
> 
> Chengwen Feng (1):
>   net/hns3: replace memory barrier with data dependency order
> 
> Lijun Ou (2):
>   net/hns3: remove unnecessary input parameter valid judgement
>   net/hns3: remove unnecessary branch
> 
> Wei Hu (Xavier) (1):
>   doc: update release notes for hns3
> 

Series applied to dpdk-next-net/master, thanks.

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2020-01-17 10:15 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-01-16  9:27 [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Wei Hu (Xavier)
2020-01-16  9:27 ` [dpdk-dev] [PATCH 1/4] net/hns3: replace memory barrier with data dependency order Wei Hu (Xavier)
2020-01-16  9:27 ` [dpdk-dev] [PATCH 2/4] net/hns3: remove unnecessary input parameter valid judgement Wei Hu (Xavier)
2020-01-16  9:27 ` [dpdk-dev] [PATCH 3/4] net/hns3: remove unnecessary branch Wei Hu (Xavier)
2020-01-16  9:27 ` [dpdk-dev] [PATCH 4/4] doc: update release notes for hns3 Wei Hu (Xavier)
2020-01-17 10:15 ` [dpdk-dev] [PATCH 0/4] improvement and cleanup for hns3 PMD driver Ferruh Yigit

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).